James Smith

But at this stage, I think that's quite unlikely.

If you ask the synthetic biologists, most of them who work on this, they'll say something like 10 to 30 years away.

But I don't think that's really pricing in the potential for transformative AI to speed things up significantly.

Another way to think about this is how much would it cost to make mirror life from where we are today?

And there people estimate something like 500 million to a billion dollars would be sufficient.

And that could make it happen much more quickly.

And there are people like George Church who think this might happen literally in the next couple of years.

So there's a lot of diversity here.

I think it's really hard to be confident in any particular estimate.

And I think that motivates the need to have these discussions now about what we should do to address the risk.

Yeah, I got involved in this middle of last year.

Before that, I was working on AI-enabled biology, doing some little bits on AI safety and actually thinking that I would work on AI safety in the longer term.

Yeah, but within two weeks of hearing about this issue, I quit the other work I was doing to focus full-time on this and have been doing that ever since.

And the main reasons for that were the tractability and neglectedness in combination with the scale of this problem.

So when I first got involved in the project, I worked with John Glass and Jack Shostak, two leading, really top scientists, to coordinate this working group of scientists and write the paper that brought this risk to the attention of the world.

Yeah, I was definitely skeptical initially.

I think pretty much everyone who first hears about this is.

And I think that's understandable.

Like, base rate, some crazy science thing, people say it's risky.

It's going to end the world.